基于顽健话者识别的矿井安全系统设计

为提高井下矿工的安全系数,加强矿井计算机系统的保密防范,便于及时获取井下矿工和计算机系统使用者的身份,用顽健话者识别设计一个易于搭建的矿井安全系统。在基于梅尔倒谱系数,与文本无关的话者识别系统上,使用基于负熵极大的快速独立量分析(ICA)实现矿井噪声的分离,对分离后的语音使用子空间语音增强以提高识别率。Matlab实验结果表明:对于联合独立量分析与子空间增强的新方法能明显优化单纯子空间增强的消噪效果,可以提高矢量量化(VQ)识别系统的鲁棒性。

Design of Mine Security System Based on Robust Speaker Recognition

In order to enhance the safety of miners and the success rate of rescue,as well as the security of computer system,a system of speaker identification is designed to build a reliable mine security policy.Using Mel Frequency Cepstrum Coefficient(MFCC),the text-independent speaker recognition system takes Fast Independent Component Analysis(ICA) algorithm based on maximum negentropy to separate the voice from noise,then allied with subspace enhancement.For Vector Quantization(VQ) system,the united way provides...